RTEMS+Cexp+GeSys

Fri May 4 14:38:08 UTC 2018

Sorry, I just found that Heinz actually also replied to this back in
december, but in the digest form. :)  **slowly working my way through
email backlog for spring semester /sigh/*

On Fri, May 4, 2018 at 10:36 AM, Gedare Bloom <gedare at rtems.org> wrote:
> Hello Anton,
>
> On Wed, Dec 13, 2017 at 9:13 PM, Anton Derbenev <aaderbenev at gmail.com> wrote:
>> Hello all,
>>
>> at NSLS-II, we have several dozen MVME3100 VME crates running RTEMS with
>> EPICS driver for various diagnostics and MRF timing hardware. The entire
>> setup was handily created quite some time ago by combining RTEMS 4.9 with
>> Cexp and GeSys to provide convenient booting and debugging capabilities. As
>> an effort to make the setup more up-to-date, I made an attempt to compose a
>> how-to which would describe how to successfully compile a bootable
>> RTEMS+Cexp+GeSys image. A nice "bonus" would be if that image also worked
>> when being run on a crate.
>>
>> It is not my first attempt as I already tried to do a similar thing for
>> RTEMS 4.10. While in both cases I found lots of work being put by the
>> community in helpful toolsets and documentation, the building process was
>> not as straightforward as one would imagine by reading the QuickStart guide.
>> There are many stumbles which, while probably being primitive in most cases,
>> could be annoying to a user who lacks core developer knowledge and
>> perspective. Some things can be mentioned _somewhere_ and can only be found
>> quickly by google-fu and/or luck.
>>
>> This time I reached for RTEMS 5 right away. Results which I achieved are
>> very similar to those which I had for RTEMS 4.10 - I can create a bootable
>> image which appears to have working GeSys and Cexp shell, but access to
>> "st.sys" fails along with other suspicious symptoms.
>>
>> Below is my "how-to" process which yields a bootable "rtems.ralf" file, and
>> results for our test MVME3100 system boot procedure. Some points highlighted
>> will probably be useful for further users' reference, and some of them come
>> from earlier conversations on the mailing list. My hope is that someone
>> knowledgeable will kindly review my steps and give some hint why the image
>> fails to work as expected. Perhaps it would be great to have a step-by-step
>> how-to as a part of documentation somewhere.
>>
>> ====================
>> RTEMS 5 + Cexp + GeSys How-To
>>
>> 1. Get RTEMS source builder (rsb). As a part of its configuration, rsb will
>> try to fetch various sources from multiple locations. If rsb is being
>> configured in a protected environment (e.g. http proxy, firewalls, etc.),
>> source locations can be unavailable. Get rsb sources on a machine which can
>> freely download from git.rtems.org, github.com etc. with both http(s) and
>> git protocol:
>>
>> $ mkdir --parents ~/src/RTEMS
>> $ cd ~/src/RTEMS
>> $ git clone http://github.com/RTEMS/rtems-source-builder.git rsb
>> $ cd rsb/rtems
>>
>> Perform dry configuration and only download sources, as suggested here:
>>
>> https://lists.rtems.org/pipermail/users/2017-September/031647.html
>>
>> Choose profile as appropriate (rtems-powerpc for MVME3100):
>>
>> $ ../source-builder/sb-set-builder --dry-run --with-download
>> --without-error-report --without-release-url 5/rtems-powerpc
>>
>> 2. If firewalls and proxies are not a problem for you, skip this step. Even
>> though sources were pre-downloaded, rsb will try to fetch the rtems-tools
>> repository during the build process (even if --no-download is specified). We
>> need to make sure the repository is accessible or the lengthy build process
>> (~ hour on two cores) will fail. In our environment, git protocol didn't
>> work through http(s) proxy, so git config was modified:
>>
>> $ cd ~/src/RTEMS/rsb/rtems
>> $ nano sources/git/rtems-tools.git/.git/config
>>
>> Replace:
>> url = git://git.rtems.org/rtems-tools.git
>> with:
>> url = https://github.com/RTEMS/rtems-tools.git
>>
>> 3. rsb now has sources and patches downloaded. If you will build RTEMS on a
>> different host, move the entire folder to the build machine.
>>
>> My build setup:
>>
>> $ uname -a
>> Linux sandbox8 3.16.0-4-amd64 #1 SMP Debian 3.16.43-2+deb8u2 (2017-06-26)
>> x86_64 GNU/Linux
>> $ cat /etc/debian_version
>> 8.9
>>
>> All further steps are performed on the build machine.
>>
>> 4. Run sb-check and install necessary packages if the check tool complains:
>>
>> $ cd ~/src/RTEMS/rsb
>> $ ./source-builder/sb-check
>>
>> Alas, sb-check doesn't check for everything. For example, the build process
>> can fail with something like:
>>
>> checking for python... /usr/bin/python
>> checking for python2.7... no
>>
>> This is resolved by being lucky in finding a similar issue discussed on the
>> mailing list. Additional packages may be needed, in the mentioned case it
>> was:
>>
>> $ sudo apt-get install python2.7-dev
>>
>> 5. Build rsb (takes a while, was an hour on my dual-core VM). Since sources
>> were pre-downloaded, specify "--no-download" (however rtems-tools will still
>> be fetched!):
>>
>> $ ../source-builder/sb-set-builder --no-download --prefix=$HOME/src/RTEMS/5
>> 5/rtems-powerpc
>>
>> 6. Build rtems. First make sure rsb tools are in path:
>>
>> $ export PATH=$HOME/src/RTEMS/5/bin:$PATH
>>
>> Get RTEMS source:
>>
>> $ cd ~/src/RTEMS
>> $ mkdir kernel
>> $ cd kernel
>> $ git clone https://github.com/RTEMS/rtems.git
>>
>> Run bootstrap scripts to generate files which are missing in cloned
>> repository:
>>
>> $ cd rtems
>> $ ./bootstrap -c
>> $ ./bootstrap -p
>> $ ../../rsb/source-builder/sb-bootstrap
>> $ cd ..
>>
>> Configure and build RTEMS for the target:
>>
>> $ mkdir build-mvme3100
>> $ cd build-mvme3100
>> $ ../rtems/configure --enable-networking --enable-cxx --enable-rdbg
>> --enable-posix --prefix=$HOME/src/RTEMS/5 --target=powerpc-rtems5
>> --enable-rtemsbsp=mvme3100
>> $ make -j 8
>> $ make install
>>
>> 7. Build Cexp. Two things which are not straightforward is getting all
>> submodules and running "make prep".
>>
>> $ cd ~/src/RTEMS
>> $ mkdir cexp
>> $ cd cexp
>> $ git clone https://github.com/till-s/cexpsh.git
>> $ cd cexpsh
>> $ git submodule update --init --recursive
>>
>> As with RTEMS bootstrap, we need to generate some missing files. Stunningly,
>> I didn't find it mentioned in cexpsh/INSTALL:
>>
>> $ make prep
>>
>> Now the code is ready to be configured and built. Use
>> "--enable-std-rtems-installdirs" to perform installation on top of the
>> previously built RTEMS tree:
>>
>> $ cd ..
>> $ mkdir build-mvme3100
>> $ cd build-mvme3100
>> $ ../cexpsh/configure --enable-std-rtems-installdirs
>> --with-rtems-top=$HOME/src/RTEMS/5 --host=powerpc-rtems5
>> --enable-rtemsbsp=mvme3100
>> $ make -j 8
>> $ make install
>>
>> 8. Build GeSys. Another trickery here - before configuring, it is necessary
>> to build the ldep tool as mentioned in documentation.
>>
>> Get all GeSys sources and submodules:
>>
>> $ cd ~/src/RTEMS
>> $ mkdir gesys
>> $ cd gesys
>> $ git clone https://github.com/till-s/rtems-gesys.git
>> $ cd rtems-gesys
>> $ git submodule update --init --recursive
>>
>> As for other parts, generate missing files by running the bootstrap tool:
>>
>> $ ./bootstrap
>>
>> Another bootstrap is necessary for ldep:
>>
>> $ cd ldep
>> $ ./bootstrap
>> $ mkdir build
>> $ cd build
>>
>> Configure and install ldep to where all rsb stuff is:
>>
>> $ ../configure --prefix=$HOME/src/RTEMS/5
>> $ make
>> $ make install
>>
>> Configure and install GeSys in the same tree where RTEMS and Cexp are by
>> using "--enable-std-rtems-installdirs". Note that $HOME/src/RTEMS/5/bin must
>> be in PATH:
>>
>> $ cd ~/src/RTEMS/gesys
>> $ mkdir build-mvme3100
>> $ cd build-mvme3100
>> $ ../rtems-gesys/configure --enable-std-rtems-installdirs
>> --with-rtems-top=$HOME/src/RTEMS/5 --host=powerpc-rtems5
>> --enable-rtemsbsp=mvme3100
>> $ make -j 8
>> $ make install
>>
>> 9. Behold the build binaries:
>>
>> $ cd ~/src/RTEMS/5/powerpc-rtems5/bin
>> $ ls
>> ar  ld      nm       objdump  readelf    rtems.ralf  st.sys
>> as  ld.bfd  objcopy  ranlib   rtems.exe  strip
>>
>> rtems.ralf is the thing. It worth mentioning that none of .obj files loaded
>> in the "st.sys" are present (telnetd.obj, monitor.obj etc.)
>>
>> ====================
>> Boot the image
>>
>> I put the resulting image in our VME timing test setup which successfully
>> boots and runs RTEMS 4.9. Boot procedure is DHCP with MOTLoad netBoot.
>> "rtems.ralf" was renamed to "rtems.boot" for booting purposes. Boot
>> sequence:
>>
>> ********************
>> Copyright(C)2008-2013,Emerson Network Power-Embedded Computing,Inc.
>> All Rights Reserved
>> Copyright Motorola Inc. 1999-2007, All Rights Reserved
>> MOTLoad RTOS Version 2.0,  PAL Version 1.2 RM04
>> Wed Nov  6 17:17:43 MST 2013
>>
>> MPU-Type             =MPC8540
>> MPU-Int Clock Speed  =666MHz
>> MPU-CCB Clock Speed  =333MHz
>> MPU-DDR Clock Speed  =166MHz
>> MPU-PCI Clock Speed  =66MHz, PCI, 64-bit
>> MPU-Int Cache(L2) Enabled, 256KB, L2CTL =A8000300
>>
>> Reset/Boot Vector    =Flash0
>>
>> Local Memory Found   =10000000 (&268435456)
>> User Download Buffer =0159D000:0179CFFF
>>
>> MVME3100> bootmem=malloc 0x500000
>> return = 0193D000 (&26464256)
>> errno  = 00000000
>> MVME3100> netBoot -d/dev/enet0 -v -abootmem -p -h -u
>> Network Loading from: /dev/enet0
>>
>> Locating BOOTP Server... Found It!
>>
>> Client IP Address      = 10.0.1.104
>> Server IP Address      = 10.0.1.1
>> Gateway IP Address     = 0.0.0.0
>> Subnet IP Address Mask = 255.255.255.0
>> Boot File Name         = rtems-5/rtems.boot
>> Load Address           = 0193D000
>> Buffer Size = ffffffff
>>
>> Network Boot File Load Start - Press <ESC> to Bypass, <SPC> to Continue
>>
>> Network Boot File Load in Progress - Press <CTRL-C> to Abort[]>>>>
>> ...
>> Bytes Received =&4886264, Bytes Loaded =&4886264
>> Bytes/Second   =&488626, Elapsed Time =10 Second(s)
>>
>> Execution Halted on Request
>> MV-----------------------------------------
>> Welcome to rtems-5.0.0 (PowerPC/Generic (no FPU)/mvme3100)
>> BSP: mvme3100, CVS Release ($Name$)
>> CPU 0x8020 - rev 0x20
>> Additionnal boot options are H
>> Initial system stack at 14c9980
>> Software IRQ stack starts at 14c9a90 with size 16384
>> Going to start PCI buses scanning and initialization
>> Board Type: MVME3100-1152 (S/N E10EA0D)
>> External (=PCI Bus) Clock Freq   :   66666666 Hz
>> Core Complex Bus (CCB) Clock Freq:  333333330 Hz
>> CPU Clock Freq:                     133333332 Hz
>> Ethernet 0                  EC:9E:CD:0E:16:5B
>> Ethernet 1                  EC:9E:CD:0E:16:5C
>> Ethernet 2                  EC:9E:CD:0E:16:5D
>> Number of PCI buses found is : 2
>> BUS:SLOT:FUN  VENDOR-DEV_ID: COMMAND STATUS BASE_ADDR0 BASE_ADDR1 IRQ_PIN ->
>> IRQ_LINE
>>   0:0x00:0    0x1057-0x0008:  0x0006 0x20b0 0x80000000 0x00000000       0 ->
>> 0 (=0x00)
>>   0:0x11:0    0x10e3-0x0148:  0x0146 0x02b0 0x80100004 0x00000000       1 ->
>> 0 (=0x00)
>>   0:0x12:0    0x10b5-0x6520:  0x0147 0x02b0 0x00000000 0x00000000       0 ->
>> 0 (=0x00)
>>   0:0x14:0    0x1095-0x3124:  0x01c7 0x02b0 0x80300004 0x00000000       1 ->
>> 2 (=0x02)
>>   1:0x00:0    0x10b5-0x9030:  0x0142 0x0290 0x80200000 0x00000000       1 ->
>> 4 (=0x04)
>> Memory:                             268435456 bytes
>> Cleared PCI errors: pci_stat was 0x20b0
>> OpenPIC Version 1.2 (1 CPUs and 56 IRQ sources) at 0x3775135744
>> OpenPIC Vendor 0 (Unknown), Device 0 (Unknown), Stepping 0
>> OpenPIC timer frequency is 41666666 Hz
>> MSR is 0x00001000, TCR 0x04000000
>> -----------------------------------------
>> Exit from bspstart
>> libi2c: Library not initialized
>> Registering mpc8540 i2c bus driver :No such file or directory
>> Tundra Tsi148 PCI-VME bridge detected at 0x80100000, IRQ 0
>> Tsi148 Outbound Ports:
>> Port  VME-Addr   Size       PCI-Adrs   Mode:
>> 0:    0x20000000 0x0e000000 0xc0000000 A32, SUP, D32, SCT
>> 1:    0x00000000 0x00ff0000 0xcf000000 A24, SUP, D32, SCT
>> 2:    0x00000000 0x00010000 0xcfff0000 A16, SUP, D32, SCT
>> 7:    0x00000000 0x01000000 0xce000000 CSR, SUP, D32, SCT
>> Tsi148 Inbound Ports:
>> Port  VME-Addr   Size       PCI-Adrs   Mode:
>> 0:    0xc0000000 0x10000000 0x00000000 A32, PGM, DAT, SUP, USR, MBLT, BLT
>> vmeTsi148 IRQ manager: looking for registers on VME...
>> Trying to find CSR on VME...
>> vmeTsi148 - IRQ manager using VME CSR to flush FIFO
>> Registering /dev/console as minor 0 (==/dev/ttyS0)
>> Welcome to RTEMS 5.0.0 GeSys
>> This system  was built on 20171213EST19:00:39
>> Installing TIOCGWINSZ line discipline: ok.
>> To skip initialization, press a key now...
>> bootpc_init: using network interface 'tse1'
>> bootpc hw address is ec:9e:cd:e:16:5b
>> My ip address is 10 .0 .1 .104
>> Time Server is 10 .0 .1 .1
>> Domain name is cs.nsls2.local
>> Boot file is rtems-5/rtems.boot
>> Command line is IOCNAME=timing
>> Subnet mask is 255 .255 .255 .0
>> Server ip address is 10 .0 .1 .1
>> Gateway ip address is 0 .0 .0 .0
>> Log server ip address is 10 .0 .1 .1
>> bootpc_adjust_interface: add net route, error=114
>> Trying to synchronize NTP...OK
>> Trying symfile 'BUILTIN', system script 'st.sys'
>> 'st.sys':
>> opening script file: No such file or directory
>> Type 'cexpsh.help()' for help (no quotes)
>> ********************
>>
>> There are several suspicious points: PCI buses scanning not reporting
>> anything, i2c library/driver errors, bootpc_adjust_interface error, and of
>> course "st.sys" script not being found. Peculiar enough, I can perform NFS
>> mount and run a script from there. It doesn't work smoothly however and
>> nfsMountsShow output is odd:
>>
>> Cexp>nfsInit(0,0)
>> RTEMS-NFS, Till Straumann, Stanford/SLAC/SSRL 2002, See LICENSE file for
>> licensing info.
>> 0x00000000 (0)
>> Cexp>nfsMount("10.0.1.1","/srv/tftp/rtems-5","/common")
>> Trying to mount 10.0.1.1:/srv/tftp/rtems-5 on /common
>> RTEMS-RPCIOD, Till Straumann, Stanford/SLAC/SSRL 2002, See LICENSE file for
>> licensing info.
>> 0x00000000 (0)
>> Cexp>nfsMountsShow(0)
>> Currently Mounted NFS:
>> 10.0.1.1:/srv/tftp/rtems-5 on /
>> 0x00000000 (0)
>>
>> ====================
>> Conclusion
>>
>> To conclude that, I'd be happy to hear any suggestions and insights to the
>> procedure described. It might be something as obvious as missing an option
>> somewhere. It would be also great if anyone who has a working
>> RTEMS+Cexp+GeSys setup would share instructions on how to make it work
>> properly.
>>
>
> I don't know if you ever worked this out for yourself. libnetworking
> does not get a lot of usage in RTEMS-5.0. Heinz Junkes tells me that
> he got EPICS-7 working on RTEMS-5 with PPC/MVME6100, but I don't know
> if they use the cexp/gesys modules yet.
>
> Gedare
>
>> Best regards,
>> Anton A. Derbenev
>> NSLS-II
>> Brookhaven National Laboratory
>>
>>
>> _______________________________________________
>> users mailing list
>> users at rtems.org
>> http://lists.rtems.org/mailman/listinfo/users